Hvdrazine
frbm Ammonia work was conducted to elucidate the processes occurring i n a high frequency discharge through ammonia a n d the conditions greatest for hydrazine yields. Hydrazine formation from ammonia in a n electric discharge (2) and during photodecomposition of ammonia has been studied (3,6 , 70).
T H I S
Experimental Procedure. A borosilicate glass flow system with a differential-pressure flowmeter at the inlet and a n absolute manometer near the discharge tube was used. .4nhydrous ammonia gas having less than 0.57, impurities passed from the floIvmeter through the spherical discharge tube (58 cc.) which contained a pair of solid nickel electrodes placed normal to the gas flow. These electrodes were cylindrical a t the base, which had a diameter of 3, 8 inch, and tapered to a point (where the discharge occurred) with a solid angle of 18’. T h e gap distance was measured with a cathetometer. From the discharge tube the gas passed through a sulfuric acid solution to the fore pump. Hydrazine was determined by passing the exit gases from the discharge tube through a 2070 sulfuric acid solution for 4 minutes, neutralizing the solution with sodium hydroxide to p H 7.2 to 7.4, and titrating it in a nitrogen atmosphere with standard iodine solution (72). Arro\$root cornstarch was the indicator. T h e extent of decomposition of the ammonia gas was determined by acid titration of the exit gas from the discharge tube for ammonia. Poiver to the discharge \vas supplied by a U. S. Army aircraft transmitter, Tvpe BC 375-E, Lvith a rated output of 150 \\atts. This transmitter has a frequency range of 1.5 to 12.5 Mc. and was operated under continuous wave conditions. Current in the discharge was controlled by tuning the antenna circuit in the transmitter. T h e po\+er to the discharge tube was fed through 6 feet of shielded RG8iU cable; ihe antenna post and ground on the transmitter were
W. H. ANDERSEN,‘ B. J. ZWOLINSK1,2 and R. 8. PARLIN3 Department of Chemistry, University of Utah, Salt Lake City, Utah
High Frequency flecfric Discharge Method The same reactions control the hydrazine formation during both the electric discharge and photodecomposition of ammonia used for connections. Stray radiation was negligible below 4 M c . per second; above this value it was troublesome. MEASUREMENT OF APPARENTPOWER DISSIPATED IN DISCHARGE. Radio-frequency probes were designed to rectify the voltages a t the discharge tube, so that they might be read directly on a direct current vacuum-tube voltmeter. These probes in conjunction with the voltmeter are peak reading, and, if the wave forms are sinusoidal. they give root mean square voltages directly. O n e probe rectifies the voltage across the discharge; the other, the drop across a noninductive resistance (100 ohms) in series with the discharge. From the latter the current is obtained. T h e product of the voltage and current gives a measure of the power dissipated in the discharge (volt-amperes). All graphs in this article which involve energy are based upon calculation of the power in this manner. T h e validity of this calculation was checked by comparing this voltagecurrent measux-ement with that deter-
Table 1.
mined by a different scherne: A portion of the discharge voltage was fed through a voltage divider (Figure 1) into tht: external synchronization circuit of a Dumont (Type 303-A) oscillograph and the current and voltage wave forms wew sequentially placed on the vertical deflection plates. Photographs (Figurr. 2) of the current and voltage wave forms (superimposed for graphical integration) were taken, the synchronization voltagt keeping the wave forms in proper phase. T h e shape of the current wave form is a characteristic produced by the apparatus rather than by the gas or discharge conditions. T h e power factors in typical cases were 0.9 and greater. The true power dissipated in the discharge probably did not deviate appreciably from that measured directly. Results. Figures 3 through 6 show the influence of various parameters describing the discharge on the energy yield of the hydrazine produced. Under the experimental conditions, maximum yields were 4 to 6 grams of hydrazine per kilowatt-hour of energy dissipation
Ammonia Conversion to Hydrazine Is a Function of Flow Rate
(Pressure 91 mm.; current 100 ma.: frequency 3.0 l l c . : gap distance 5.1 mm.) %h”3 Molar % Decomp. Molar % Total Cc./Min. (STP) Decomp. SH3 t o N?H. N H , to Ni?H, Flow Rate,
300
5.7 5.0 3.4 2.8 1.6 0.9
400 600 800 1000 1200 1400
0.265 0.329 0.334 0.317 0.299 0.273 0.257
4.7 6.6 9.8 11.3 18.6 30.4 37.1
0.5
n
Present address, Aerojet-General Corp.,
Azusa, Calif.
* Present
address, Carnegie Institute of Technology, Pittsburgh, Pa. Present address, Operations Research, Inc., Silver Spring, Md.
Radio-frequency probes rectified voltages for direct reading 1.
5.
1 00-ohm resistor Flashlight battery
2.
Flashlight case
3.
Condenser, 0.02 mfd.
4.
VOL. 51; NO. 4
Rectifying tube, 1 B3GT
APRIL 1959
527
TO EXTERNAL
DISCHARGE
47$
* I
TO VERTICAL DEFLECTION PLATES ON SCOPE
SYNCH.
#
100 K
-
,'
Figure 1. A voltage divider circuit was used to obtain the phase angle between the voltage and current in the discharge
in the discharge. These 1-ields could be increased, if all the discharge parameters were maintained for maximum yield. Yields \vere not appreciably changed by changing the frequency through the range of 1 . 5 to 7.0 Mc. per second. Above 7 I r k . the discharge could no longer be maintained. Immersing the discharge tube in constant temperature baths of -23', 0":and 50" C. had no influence on yields. T h e discharge tube normally reached 90' to 150' C. T h e influence of the discharge parameters on the rate of decomposition of ammonia is shonm in Figure 7. T h e fraction of decomposed ammonia converted to hydrazine increases rapidly as the floiv rate is increased (Table I).
Figure 2. Individual current (upper) and voltage (lower) wave forms in 3Mc. discharge in ammonia gas
The fraction of the total ammonia converted to hydrazine goes through a maximum. however, and is very sma!l. Under optimum conditions of lolver gas pressure, 1o\+er discharge current, and shorter gap distance, the fraction of decomposed ammonia converted to hydrazine undoubtedly is higher than that shown in Table I.
Discussion Primary Processes. Because an electron is small, it will store much of the
0.3. FRE9UENCY 3.0 M C G A P DISTANCE 5.1 MM. CONTACT TIME e J X 1 6 S E C
0.3
0.2GIHI
0.1-
0 0
100
" 400
300
205
-
PRESSURE I N M M HO
Figure 5 FREQUENCY F L O W R A T E IN C C l M l N
Figure 3
F R E Q U E N C Y 3.OM C I FLOW R A T E 390 CC/Mlh GAP D I S T A N C E 5.1 M M .
03t 59MM
CURRENT
0.2 -
c""
3.0 MC. 100 MA.
- \
FLOW RATE 390 C.C/MIN
k
0.2 -
i(n)
01
O
. 50
100 150 CURRENT IN M I L L I A M P E R E S
Figure 4 Figures 3 to 6.
4
200
12
Figure 6 Effect of parameters on hydrazine formation
Yields can b e increased by regulation of all discharge parameters. formed per 100 e.v. of energy dissipated in the discharge
528
8
GAP DISTANCE I N M M
INDUSTRIAL AND ENGINEERING CHEMISTRY
G(H) is the molecules of hydrazine
energy it acquires from an electric field through many collisions Lvith gas molecules until a nonelastic collision occurs. Its near-elastic collisional energy losses are due primarily to molecular vibrational and rotational excitation ( 9 ) . For some molecules subjected to electron impact essentially only one path leads to decomposition Lvithin a restricted electron collisional energy range --i.e., one predominant set of initial decomposition products is usually produced. Mass spectra data of ammonia (8)show little probability that a n electron with less energy than the ionization potential (10.5 e.v.) can dissociate the mdecule and give a n ion as one prodnct. Little is known about neutral products from lo\ver energy electron impact. -4simple model is offered as a guide for estimating effects of discharge parameters on the rate of ammonia decomposition. The electron energy distribution is approximated to be Maxwellian. This is reasonable in view of the relatively large calculated electron density. ca. elrctrons per cc. ( 3 ) . The distribution of electron energiis due to the Eeld is superimposed on the thermal energy of the electrons. This thermal energy is ignored, as it is very small. The molecules and ions possess a lower energy than the electrons; thus. the electrons are not in thermal equilibrium with the gas molecules or ions. I t is now assumed that a n electron on the average will travel S cm. in the field direction before it accumulates enough energy to decompose a molecule through a n inelastic collision (excitation or ionization). At the end of the inelastic impact the electron is assumed to lose all its energy and must begin again. Thus, unless a n electron gains a n energy Xesi >, E in a path of length Si>it cannot cause dissociation. E is the activation energy for the process which leads to dissociation of the molecule, X is the field strength, and e is the electronic charge. Electronmolecule collisions with energy greater than E will probably leave one or more dissociation products in a n excited state.
H Y D R A Z I N E F R O M AMMONIA 400
800
1
I
FREQUENCY X
' X ' P R E S S U R E , M. H g
FLOW RATE,CC/M
1
I
G A P 9ISTANCE 5.1 MM
3.0 M C
PRESSURE CURVE
1800
1200
I O O M A , 3 3 0 CC/MIN
A CURRENT C U R V E . 91 MM, 3 9 0 CC/MIN 0 FLOW R A T E C U R V E . 91 M M , 100 M A
/
../
'
I GAP
v
2
50
I
100
IO0
I50 200
T h e chance per electron of producing a dissociative collision is dependent on the chance of a n accumulative electron free path S,2 S = E ' X e in ihe direction of the field. O u t of .\-o electrons starting out per second, the number .I7per second that have paths Si> S are assumed to be of the same form as the Maxwellian distribution law for free paths, with S replacing the free path-i.e., o'\I
exp ( - S / S i )
x
MA
300 PRESSURE, M I
Figure 7. The fraction of ammonia influenced by the discharge parameters
S =
, CURRENT,
DISTANCE 5.1 M M , ~
(1 1
This model is thus someivhat analogous to that first used by T o ~ i n s e n dto relate the primary ionization coefficient to the discharge parameters. Loeb has given a detailed criticism ( 3 ) . T h e treatment below circumvents certain difficulties inherent in the Townsend model. Moreover. primary chemical effects (dissociation) induced by electron collisions in a low energy discharge will be less sensitive to the precise electron distribution than ionization characteristics. S = E Xe, and Si may be Xvritten approximately as nL. \\here I, is the component of the electron mean free path parallel to the field, and n is a number which Lvhen multiplied into I, gives Si. T h e value of n is perhaps a Lveak function of X,'P. Holvever, in the present treatment n is considered constant. This is justified for t\vo reasons: T h e X P range is relatively small (Table 11) and the functionality of n in the final rate equation (Equation 2) leaves the rate virtually insensitive to small changes in n. L should vary inversely a s the pressure and can be written as L,(760 P):where L, is L a t 1 atm.: and P i s the absolute gas pressure in millimeters of mercury. T h e value of may be transformed to a molecular decomposition rate as follo\vs: T h e total (effective) number of electrons per second, .Yo>is I X 6.24 X 10'8, where I is the average discharge current in amperes. If the electron density is small compared to the gas density: the number
decomposed
is
$0'GAP
DISTANCE, MM
Figure 8. Electron and voltage characteristics of discharge are similar to those in a low pressure direct current discharge
of electron-molecule collisions per electron M hich lead to molecular decomposition is I nL. I\ here I is the average distance a n electron travels in its lifetime. T h e final expression for the rate of decomposition of gas molecules. 'M per second, through \vhich a n electron m a r m is moving under the influence of a n electric Eeld can be Lvritten
IPI(6 24 X lo'*) - d.M - - fi@ exp dt 760 nL,
r
-E
i
v here k; is the transmission coefficient and is related to the conventional quantum yield and accounts for t h r possible reformation of original gas molecules from secondary reactions. It is a function mainly of the gas flolv rate through the discharge. I n the frequency range usrd. an electron undergoes many collisions betxveen cycles of the applied field. Furthermore, the electron drift time across the discharge is short compared to the field reversal time. T h e current and field strength are thus essentially unidirectional when the electrons are being acted upon by the field between collisions. T h e electron and voltage characteristics of the present discharge (Figure 8) are therefore similar to those in a low pres sure direct current discharge, and during
Table II.
a given half-cycle: the zones in the discharge are probably similar to those in the direct current case. Small calculated differences are due to field distortions from a positive space charge built u p in the high frequency discharge from trapped positive ions. T h e discharge region thus consists roughly of the cathode drop region characterized by a large potential gradient and a short distance, and the positive column region, with a low voltage gradient and a r d a tively great distance. The current in the positive column is principall?. a n electron current; in the cathode drop region, partially positive ions. T h e total rate of gas decomposition in the discharge should therefore be given approximately by the sum of two expressions (Equation 2): corresponding to the tlvo discharge regions. Because. of difficulty in estimating 1 and I for the cathode drop region in the present investigation, this calculation was not practical. Instead, the approximation was made that the effective field strength is given by the average field strength of the discharge-i.e., X = I' d. \vhere I; is the discharge voltage and d is the electrode gap distance. T h e rate of gas decomposition in the discharge is then given by Equation 2: the average field strength being used. I n the relatively high pressure range used almost all current flow is confined to the discharge.
Decomposition of NHI in Electric Discharge Is Linear
Pressure, Mm.
Total Discharge Voltage
Total Discharge,
39 69 91 140 276
555 560 565 575 640
27.9 15.9 12.2 8.0 4.6
x/p
Mole NH3 I)ecomp.,'Sec. __________
Normalized
2.9 10.1 14.5 22.6 39.2
X
x
10-5
X X X
VOL. 51, NO. 4
17.1 16.3 14.5 13.0 10.0
X
x
10-6
X X X
APRIL 1959
529
T h e value of I in Equation 2 is therefore approximated to be d. T h e decomposition data a t various pressures (Figure 7 ) were normalized to the same quantity of substrate and equal residence time in the discharge as occurs a t 91 mm. (an arbitrary value), by multiplying the percentage of ammonia decomposed a t each pressure P by the factor (91/P), and the factor %D(390) ’7oD(390 X 91 /P), where 700( ) is the percentage decomposed a t a mass velocity of 390 and 390 X 91/P cc. per minute (Table 11). These values were obtained from the flow rate curve. A plot of the logarithm of rate of ammonia decomposition as a function of (X, P)-‘gives a reasonably straight line. I n applying Equation 2 X was taken as unity and @ was calculated to be 0.6 by extrapolating decomposition as a function of flow rate (Figure 7) to zero flow rate and forming the ratio of the decomposition a t 390 cc. per minute to that a t zero flow rate. From the ordinate intercept of the plot, n was calculated to be 6350. This, with the slope of the line, leads to a calculated E of about 4.2 e.v. This indicates that primary decomposition takes place for the most partwithout ionization. This agrees with conclusions (2, 7 7 ) from other considerations. It suggests that principally only the excitational decomposition mechanism NH3 -w+
NHz
+H
(3)
which requires a n activation energy of about 4.5 e.v., occurs. Electron-molecule collisions with energy greater than about 4.5 e.v., but less than the ionization potential (10.5 e.v.), probably produce excited NH2 radicals (74). Devins and Burton (2) conclude that two separate primary decomposition reactions take place; one leads to hydrazine formation (Equation 3), and some other mechanism leads to nitrogen formation. T h e crudeness of the present model and experimental data d o not preclude a more complex over-all decomposition scheme than the above treatment suggests. T h e decomposition path ”3
-wd
NH
+ Hz
(4)
may be possible. and if its activation energy is near that of Reaction 3, the two reactions may occur simultaneously a t nearly equal rates. T h e analytical results involving Equation 2 cannot reveal this possibility. Previous work ( 7 3 ) suggests that neutralization of the ions a t the electrodes or walls does not lead to dissociation. T h e NH3+ ions probably neutralize without dissociation. because a nonbonding orbital is involbed (7). Secondary Reactions. T h e photochemical mechanism of Gunning and others (5, 6 ) is consistent with the experimental findings. NHI -A-
530
NH2
+H
(5a)
ka + NzHd + N?HI + H Z kt H + N2H3 +NZ + 2H2
H
NIHa -9-
H
+ N3H3
+ NHz 2NHJ k 2H + Wall L+ H? H
(5~) (jd) (je) (5f)
(%I
T h e k’s represent specific rate constants. Reaction 5a is from evidence presenred above. With the pressure used in this study Reaction 5b probably occurs through a homogeneous gas reaction. Further evidence is the lack of increase in yields when the discharge tube is maintained a t low temperatures. T h e fraction of the decomposed ammonia converted to hydrazine rapidly increases a5 the flow rate is increased (Table I). Thus part of the nitrogen must come from the radical attack decomposition of hydrazine; hence Reactions 5c and 5d. Reaction 5e is included to explain Figure 4 (7). Table I shows that ammonia is regenerated by secondary reactions and suggests that hydrazine destruction does not give appreciable ammonia; thus Reaction 5f is included. T h e usual steady-state treatment for Reactions 5a to 5g gives the following expression for the rate of decomposition of ammonia to very close approximation :
kzki
[s
dx
dWFj [C - D X
+ Exz]’/* (7)
where x is the fractional conversion of ammonia to hydrazine, F is the flow rate, and the constants involve the various rate constants, values of ne, reactor volume. pressure, and averaged fractional conversion to nitrogen. T h e choice of signs arises from the square roots involved. In Figure 9 data given in Table I are plotted. Equation 7 can be separated into two equations by using the choice of algebraic signs as shown. The equations fit the experimental data reasonably well. T h e solutions of Equation 7 become imaginary above a certain value of x (the maximum in the curve). Relating the reactions of Equation 5 to the experimental data is further evidence they are consistent with experimental observations. However. this reaction scheme may be modified and still fit the experimental data. The approximations necessary for solution of more
2
3
5
4
I / F , I O - ~SEC./HOLE
Figure 9. Fractional conversion of ammonia to hydrazine is a function of flow rate
complex mechanisms preclude obtaining more detailed information.
literature Cited (1) Birse, E. A. B., Melville, H. W., Proc. Roy. SOC.(London)A175, 164 (1940). (2) Devins, J. C., Burton, M., J . Am. Chem. SOC. 76, 2618 (1954). (3) Loeb, L. B., “Fundamental Processes of Electrical Discharges in Gases,” pp. 358-68, Wiley, New York, 1939. (4) Zbid., p. 585.
+ 21 + ’klkan, (NH3)(NZH4) [& ks kzk?
where ne is the electron density. Solving Equation 6 in a manner approximate for flow systems (75) gives
INDUSTRIAL AND ENGINEERING CHEMISTRY
I
(F
(NHa))’”
+ k3(N2H4)] 1
l”
(6)
(5) McDonald, C. C., Gunning, H. E., J . Chem. Phys. 23, 532 (1955). (6) McDonald, C. C., Kahn, A., Gunning, H. E., Zbid.,22, 908 (1954). (7) McDowell, C. A., Zbid., 24, 618 (1956). (8) Mann, M: M., ‘Hustrulid, A.,. Tate, J. T., Phys. Reu. 58, 340 (1940). (9) Massey, H. S. W., Burhop, E. H. S., “Electronic and Ionic Impact Phenomena,’’ p. 279, Oxford University Press, 1952. (10) Noyes, W. A., Leighton, F. A., “Photochemistry of Gases,” pp. 370-82, Reinhold, New York, 1941. (11) Ouchi, K., Takamatsu, T., J . Electrochem. Sod. Japan 21, 75, 132 (1953). (12) Penneman, R. A., Audrieth, L. P , Anal. Chem. 20, 1058 (1948). (13) Smith, C., Essex, H., J . Chem. Phys. 6, 188 (1938). (14) Terenin, .4., Neujmin, H., Zbid, 3, 436 (1935). (15) Wilde, K. A., Zwolinski, B. J., Parlin, R. B., J. Am. Chem. SOC.79, 1323 (1957). RECEIVED for review June 3, 1958 ACCEPTED November 19, 1958 Division of Physical and Inorganic Chemistry, 122nd Meeting, ACS, Atlantic City, N. J., September 1952. Abstracted from a thesis submitted by W. Hoyt Andersen to the faculty of the University of Utah in partial fulfillment of the requirements for the degree of doctor of philosophy, 1952. Work supported by the U. S. Atomic Energy Commission.